Re: [PATCH v2 2/2] mm: Use vma_start_write_killable() in dup_mmap()

[Lorenzo Stoakes <lorenzo.stoakes@oracle.com>]

On Mon, Nov 10, 2025 at 08:32:02PM +0000, Matthew Wilcox (Oracle) wrote:
> Allow waiting for the VMA write lock to be interrupted by fatal signals.
> The explicit check for fatal_signal_pending() can be removed as it
> is checked during vma_start_write_killable().  Improves the latency
> of killing the task as we do not wait for the reader to finish before
> checking for signals.

Does this fix Chris's issues or is a forthcoming series required for that?
I'm guessing not, as his issue was w.r.t. reclaim right rather than at the
point of forking?

If it does we should say so I think. If not then obviously not!

>
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>

LGTM so:

Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>

> Reviewed-by: Suren Baghdasaryan <surenb@google.com>
> Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
> ---
>  mm/mmap.c | 12 +++---------
>  1 file changed, 3 insertions(+), 9 deletions(-)
>
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 5fd3b80fda1d..03773fe777b7 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -1750,7 +1750,9 @@ __latent_entropy int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm)
>  	for_each_vma(vmi, mpnt) {
>  		struct file *file;
>
> -		vma_start_write(mpnt);
> +		retval = vma_start_write_killable(mpnt);

Maybe worth explicitly adding a comment to say we handle fatal signals
here? I guess it is implied but still maybe useful?

> +		if (retval < 0)
> +			goto loop_out;
>  		if (mpnt->vm_flags & VM_DONTCOPY) {
>  			retval = vma_iter_clear_gfp(&vmi, mpnt->vm_start,
>  						    mpnt->vm_end, GFP_KERNEL);
> @@ -1761,14 +1763,6 @@ __latent_entropy int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm)
>  			continue;
>  		}
>  		charge = 0;
> -		/*
> -		 * Don't duplicate many vmas if we've been oom-killed (for
> -		 * example)
> -		 */
> -		if (fatal_signal_pending(current)) {
> -			retval = -EINTR;
> -			goto loop_out;
> -		}
>  		if (mpnt->vm_flags & VM_ACCOUNT) {
>  			unsigned long len = vma_pages(mpnt);
>
> --
> 2.47.2
>
>